reinautils 0.0.3

General utilities

reinautils

The Utilities included are:

---

source

Parameters

 Parameters (**kargs)

A splecial class whos atributes can be referenced as attributs or as dictionaty keys

---

source

device_by_name

 device_by_name (name:str)

Return reference to cuda device by using Part of it’s name

Args: name: part of the cuda device name (shuuld be distinct)

Return: Reference to cuda device

Updated: Yuval 12/10/19

---

source

DatasetCat

 DatasetCat (*datasets)

Concatenate datasets for Pytorch dataloader

The normal pytorch implementation does it only for raws. this is a “column” implementation

Arges: datasets: list of datasets, of the same length

Updated: Yuval 12/10/2019

Install

pip install reinautils

How to use

Parameters

You can create a Parameters class from dict

params=Parameters(first=1,second='A')
print(params.first)

1

You can also creat a Parameters class and populate it from a json file

params2=Parameters().from_json('config_demo.json')
print(params2)

Parameters:
   path : Parameters:
      data : /workspace/hd/
      tmp : /workspace/hd/tmp/
      features : /workspace/nvme/features/
      train : /workspace/nvme/train/
      models : /workspace/hd/models/
      output : /workspace/hd/outputs/
      test : /workspace/nvme/test/
   platform : myserver